The present invention relates to a method for manufacturing a semiconductor thin film that can be used in short-wavelength light emitting diode devices, short-wavelength semiconductor laser devices and high-speed electronic devices.
A group III–V nitride semiconductor material, having a wide forbidden band, can be used in various light emitting devices, such as light emitting diode devices and short-wavelength semiconductor laser devices that are capable of emitting light of a color in a visible region such as blue, green or white. For its high electron saturated drift velocity, in addition to the wide forbidden band, it is also a promising material for electronic devices such as high-frequency devices and high-power devices.
Among others, light emitting diode devices using a nitride semiconductor have already been in practical use in large-size display apparatuses, traffic signals, etc. Particularly, white light emitting diode devices, which give white light by exciting a fluorescent substance with blue or ultraviolet light, can be used in lighting fixtures to replace electric bulbs and fluorescent lamps, and thus bears high expectations. On the other hand, the research and development of semiconductor laser devices has reached a point where samples are being shipped and products are being manufactured although in small quantities, aiming at application in optical disk apparatuses of larger storage capacities.
For the application in semiconductor laser devices, it is important to increase the operating lifetime. To this end, active research and development has been undertaken to reduce the crystal defects in a nitride semiconductor used in a laser device. At present, the most effective method for reducing defects is an epitaxial lateral overgrowth (ELO) method. Specifically, a thin film made of, for example, silicon oxide (SiO2), or the like, is selectively formed on a base layer made of gallium nitride (GaN) so that the formed thin film has openings therein, and the nitride semiconductor is re-grown starting from the gallium nitride surface exposed through the openings. As a result, the crystal growth in the lateral direction (the direction parallel to the substrate surface), which proceeds starting from the openings, becomes dominant on the SiO2 thin film. By promoting the lateral growth, it is possible to significantly reduce the crystal defects occurring in the nitride semiconductor layer, which is grown on the mask film. It has been reported that with this method, the dislocation density, which is about 1×109 cm−2 in the prior art, can be reduced to be on the order of 106 cm−2.
A semiconductor laser device manufactured by the ELO method has a maximum operating lifetime exceeding 10000 hours, which is substantially sufficient for practical applications. Further reduction of the defect density is expected in order to realize a reliable laser device having an even longer operating lifetime. On the other hand, it has also been reported that when a GaN selective growth layer is grown by the ELO method to a thickness exceeding about 100 μm on a sapphire substrate, the substrate is separated from the selective growth layer while the substrate is being cooled after the crystal growth process, due to the stress caused by the difference between the coefficient of thermal expansion of the substrate and that of the GaN layer.
The separation of the GaN layer from the sapphire substrate provides various advantageous effects, including: the heat radiating property is improved as compared with a case where the sapphire substrate is not separated; an electrode can be formed on the reverse side of the substrate, thereby simplifying the process; and the substrate can be cleaved, thereby realizing a desirable mirror (cavity facet) while reducing the chip size. As a result, it is possible to increase the operating lifetime and the performance of a semiconductor laser device.
A conventional method for manufacturing a semiconductor thin film using the ELO method will now be described.
FIG. 25A and FIG. 25B are cross-sectional views sequentially illustrating steps in the conventional method for manufacturing a semiconductor thin film.
First, as illustrated in FIG. 25A, a first semiconductor layer 102 made of gallium nitride (GaN) and having a thickness of about 1 μm is grown as a base layer on a substrate 101 made of sapphire by a metal organic chemical vapor deposition (MOCVD) method, for example. Then, a mask-forming film made of silicon oxide (SiO2) and having a thickness of about 200 μm is deposited on the first semiconductor layer 102 by a chemical vapor deposition (CVD) method, for example. Then, a resist pattern having a planar stripe pattern (not shown) is formed on the mask-forming film by a photolithography method, and then the mask-forming film is wet-etched with hydrogen fluoride using the formed resist pattern as a mask, thereby forming, from the mask-forming film, a mask film 103 having a stripe pattern.
Next, as illustrated in FIG. 25B, a second semiconductor layer 104 having a thickness of about 100 μm is re-grown (through an ELO process) on the first semiconductor layer 102 with the mask film 103 formed thereon by a hydride vapor phase epitaxy (HVPE) method. In this process, gallium chloride (GaCl), which is a group III source obtained by reacting metal gallium (Ga) with a hydrogen chloride (HCl) gas, and ammonia, which is a nitrogen source, are used as material gases. With the HVPE method, the growth of the second semiconductor layer 104 made of gallium nitride starts from portions of the first semiconductor layer 102 that are exposed through the mask film 103, and the growth proceeds also in the lateral direction so as to cover the top of the mask film 103. After the second semiconductor layer 104 is flattened, the growth proceeds while retaining the flat surface. Since the lateral growth portions on the mask film 103 grow without being influenced by the crystal defects of the base layer, it is possible to significantly reduce the crystal defect density.
After the second semiconductor layer 104 is grown to an intended thickness, the structure is cooled to room temperature. In the cooling process, the substrate 101 is warped into a convex shape due to the difference between the coefficient of thermal expansion of the substrate 101 made of sapphire and that of the first and second semiconductor layers 102 and 104 made of gallium nitride. As a result, the substrate 101 and the first semiconductor layer 102, or the mask film 103 and the second semiconductor layer 104, are detached from each other at the interface therebetween, thereby separating the substrate 101 from the first semiconductor layer 102 or the second semiconductor layer 104.
In the conventional method for manufacturing a semiconductor thin film, by decreasing the size of the openings of the mask film 103 as much as possible, it is possible to make the lateral growth of the second semiconductor layer 104 dominant without being influenced by the defect density of the base layer (the first semiconductor layer 102), whereby it is possible to reduce the defect density in the second semiconductor layer 104.
However, in order to reduce the opening size to a deep submicron level, for example, it is necessary to use a micro pattern exposure system and a micro etching apparatus that have high precisions. Thus, further reducing the defect density in the second semiconductor layer 104 requires high-precision tools, thereby increasing the manufacturing cost.
Moreover, it is not possible to obtain openings of a size below the lower limit, which is dictated by the level of the state-of-the-art microprocessing technology. Thus, the cost of the process and the size of the openings can only be reduced to a certain degree. As a result, there is a limit to the reduction of the crystal defect density in the second semiconductor layer 104 through the miniaturization of the pattern of the mask film 103.
Moreover, when the thickness of the second semiconductor layer 104 is increased, there is a problem occurring in the step of detaching the substrate 101. That is, it is difficult to uniformly and reproducibly obtain a semiconductor layer of a large area due to problems relating to the thickness distribution across the plane of the second semiconductor layer 104 and the reproducibility of the interface condition thereof.